haruka

I built a Discord bot that grew to 80,000+ servers. Consistently one of the highest voted on top.gg. I left that project.👇

Any benefits of using Codex app over CLI? Is it the same underlying harness?

Hmm when you think about it, do you even need commands on Discord? What if you just tell the bot in plain language what you want it to do?

@dmontetheno1 Slightly tangential but it reminds me of this x.com/oneubon4563/st… The language used can be to address specific peoples/regions

The “Syriac dominance” thesis presupposes that the Qur’an’s theological lexicon is primarily derivative of Syriac Christian usage. Suleyman Dost argues that premise is unstable when South Arabian epigraphy is treated as primary comparative evidence (1/25)

Would be great if Codex could fight back a little on my plan. Yeah I know the plan says “do this” but it should be able to go “well I think doing it like that would be better instead”

When we were releasing our bot, we had a good number of people waiting for the launch, so they could get the rarer stuff. A short while earlier, I pushed an “optimisation” that I didn’t really test. Ended up breaking things so bad we had to relaunch.

It’s still really early days. Will hope to open source soon to let you self host. I’ll be posting the dev journey here, so if you run a Discord server or you’re interested in bot dev, follow along.

Which is why I’m building GuildMate. It’s an AI powered bot built specifically for Discord, to understand and help your community. It will remember context about your members and can build new commands for itself, based on what you want.

I built a Discord bot that grew to 80,000+ servers. Consistently one of the highest voted on top.gg. I left that project.👇

Taalas came out of nowhere and actually did something huge

@4shadowed Hey I’m working on something similar, so far I’m leaning towards Cartesia Sonic (tts). They have a stt api too which I want to try but otherwise might just use whisper for that

I’m very behind, what’s the latest and greatest on super fast STT and TTS? I still just send everything off to whisper

@nick_tikhonov Hey enjoyed the write up (and glad you added voice snippets from your testing). The final result looks amazing, could for sure have a real time conversation I’m interested in extending something like this to multi speaker input. Wonder if you had any experiences/ideas?

built a SOTA voice agent from scratch in ~1 day. ended up beating off-the-shelf platforms by 2× on latency (~400ms e2e). here's the write-up 👇

>buy a gpu >run your own model locally >actually to run anything decent you’ll need many gpus >each one is expensive >dont forget electricity bills >you will most likely be behind actual sota

@karpathy Do think we are seeing the early scaffold of the new “botnet” - imagine a malicious actor at this stage

I'm being accused of overhyping the [site everyone heard too much about today already]. People's reactions varied very widely, from "how is this interesting at all" all the way to "it's so over". To add a few words beyond just memes in jest - obviously when you take a look at the activity, it's a lot of garbage - spams, scams, slop, the crypto people, highly concerning privacy/security prompt injection attacks wild west, and a lot of it is explicitly prompted and fake posts/comments designed to convert attention into ad revenue sharing. And this is clearly not the first the LLMs were put in a loop to talk to each other. So yes it's a dumpster fire and I also definitely do not recommend that people run this stuff on their computers (I ran mine in an isolated computing environment and even then I was scared), it's way too much of a wild west and you are putting your computer and private data at a high risk. That said - we have never seen this many LLM agents (150,000 atm!) wired up via a global, persistent, agent-first scratchpad. Each of these agents is fairly individually quite capable now, they have their own unique context, data, knowledge, tools, instructions, and the network of all that at this scale is simply unprecedented. This brings me again to a tweet from a few days ago "The majority of the ruff ruff is people who look at the current point and people who look at the current slope.", which imo again gets to the heart of the variance. Yes clearly it's a dumpster fire right now. But it's also true that we are well into uncharted territory with bleeding edge automations that we barely even understand individually, let alone a network there of reaching in numbers possibly into ~millions. With increasing capability and increasing proliferation, the second order effects of agent networks that share scratchpads are very difficult to anticipate. I don't really know that we are getting a coordinated "skynet" (thought it clearly type checks as early stages of a lot of AI takeoff scifi, the toddler version), but certainly what we are getting is a complete mess of a computer security nightmare at scale. We may also see all kinds of weird activity, e.g. viruses of text that spread across agents, a lot more gain of function on jailbreaks, weird attractor states, highly correlated botnet-like activity, delusions/ psychosis both agent and human, etc. It's very hard to tell, the experiment is running live. TLDR sure maybe I am "overhyping" what you see today, but I am not overhyping large networks of autonomous LLM agents in principle, that I'm pretty sure.

The last ~20% of a project is really a crunch

@arcee_ai ~~Small typo~~

Today, we’re releasing the first weights from Trinity Large, our first frontier-scale model in the Trinity MoE family.

Anyone used the Alibaba Cloud trial? Looks pretty generous

@ImSh4yy That’s a good shout, might have to keep an eye open on marketplace

Can’t wait for all these Mac Minis to hit the second hand market once the hype for whatever this is dies down.

I’ve found Gemini 3 Pro better for brainstorming ideas, compared to Opus 4.5 Opus was going too far but Gemini kept it more in the requirements Didn’t try GPT 5.2 yet, or are there any other models that are good for this?